Inper Smart AI-Based 3D Behavioral Analysis System

Track animal pose with matched models and multi-view video for live behavior events, conditional feedback and subsequent analysis.

Discuss configuration See behavior through pose. Inper Smart AI-Based 3D Behavioral Analysis System Combine multiple video views with a matched AI model to track keypoints, display pose and behavior, and connect detection conditions to experimental feedback. The chamber contains four side-view cameras and one top-view camera, adjustable lighting and ventilation. Inper One runs Smart to manage cameras, models, recordings and events. Chamber, cameras and workstation together. 02 · System The original four-side-view clip shows animals moving in the clear enclosure. Camera position, focus and aperture affect model input; preserve the configured optics and confirm the field matches the experiment. Keep several views of the same behavior. 03 · Multiple views Each model has a scope for species, animal count, camera views and behavior labels. A base rodent model is not universal; new scenes require appropriate training, configuration and validation. Match the model to the scene. 04 · Model selection Choose the mode for the acquisition task, then assign the required cameras. Record mode captures video for subsequent annotation, model preparation and analysis. Record Keep video for later work The documented 3D-M1 workflow selects and applies four side-view cameras to capture mouse keypoints and reconstruct a 3D skeleton. 3D-M1 Reconstruct pose from four side views Record video or apply a pose model. The corresponding rodent model forms a skeleton from body keypoints. Zoom and change the 3D viewpoint, checking output against camera images rather than interpreting occlusions or errors as movement. See body relationships between keypoints. 06 · Three-dimensional pose Tracking can display movement speed and body length, while Behavior shows selected labels over time. In the documented version, display checkboxes affect visibility; the corresponding data remain in the recording. Follow speed and body length over time. 07 · Live tracking parameters Select model-supported behaviors for the experiment and check detections against video. The 3D-M1 examples include walking, turning, rearing, grooming and stretching. Available labels follow the current model list. Basic behavior From movement to posture Social models can provide interaction labels such as sniffing, following and contact. Animal count and scene must match, with performance checked for colors and occlusions. Social behavior Interaction needs a matched model Behavior labels come from the model. The original social-model screen recording places video, keypoints and events together. It demonstrates that model’s workflow rather than a universal accuracy claim across species, counts or complex scenes. Review keypoints in a real recording. 09 · Multi-animal example Markers can originate from manual input, external DIO, internal device events or model Tracking/Behavior conditions. Rules such as persistence define the event and should be checked against video. Turn behavior conditions into events. 10 · Markers Output can send 5 V TTL from movement-speed or behavior conditions to mark or trigger compatible equipment. Condition windows and pulse settings are not inference-latency specifications; validate the complete setup. From a detected condition to an output signal. 11 · Conditional feedback Configure immediate, delayed, scheduled, duration-based and internal/external triggers. In multi-instrument experiments, verify inputs, device conditions and stop rules before recording. Fit acquisition timing to the task. 12 · Start and stop triggers Preview checks the model and animal without saving data. Use Start for the recording session; in the documented 3D-M1 workflow, also check keypoint overlays during formal acquisition. Check detections before recording. 13 · Preview and acquisition Config saves models, cameras, frame rate, display items and Marker/Output for reuse under the same workstation account. Recheck Trigger separately and review performance after environmental changes. Save model and cameras as the next starting point. 14 · Configuration reuse Name and tag recordings, then select them in Data and export to a drive. The documented One acquisition application does not perform offline analysis; subsequent visualization and statistics use compatible analysis software. Keep video and model output for further analysis. 15 · Data handoff Companion analysis can organize ethograms, behavior statistics and transition relationships. The source image illustrates functions; available analyses depend on exported data, the behavior model and analysis software version. Compare behavior frequencies and transitions. 16 · Behavior analysis Inspect distributions, temporal changes and condition differences in model-derived body parameters. Body parameters Companion analysis offers parameter heatmaps, densities and dimensionality-reduction views, interpreted with the parameter definitions. Heatmaps and dimensionality reduction Check unusual segments against source video and keypoints before including them in statistics. Review against source Explore several views of pose data. Define camera views, lighting, species and animal count. Define scene conditions Match an existing model where possible; new animals or labels require annotation, training and validation planning. Select or train Check recognition, data export and feedback output together through the complete experiment. Validate the workflow Model scope belongs in the plan. System Behavior chamber, 5 cameras, Inper One and Smart software Camera views 4 side views + 1 top view; 3D-M1 uses the 4 side views Frame rate Cameras share the selected rate; the manual recommends 30 fps for general behavior, subject to configuration Model scope Species, animal count, views and behavior labels must match the model Live display Video, pose skeleton, Tracking, Behavior and Event Events and feedback Marker, Trigger and conditional TTL Output Offline analysis Use compatible analysis software and models after data export Parameter Description Specifications and functions depend on the model, software version and delivered configuration. Technical specifications Inper Smart AI-Based 3D Behavioral Analysis System — User manual Inper Smart AI-Based 3D Behavioral Analysis System — Product brochure Inper 2026.B Product Catalog Downloads Share species, animal count, target behaviors, camera/chamber scene, feedback needs and analysis goals to confirm model scope and training requirements. Ask about Inper Smart AI-Based 3D Behavioral Analysis System Send inquiry Product selection Configuration and quotation Technical and application support